TorchUncertaintyCalibratedLogitRepresenter¶

class probly.integrations.torch_uncertainty.TorchUncertaintyCalibratedLogitRepresenter(predictor: Predictor[In, Out], *_args: CtrIn.args, **_kwargs: CtrIn.kwargs)[source]¶

Bases: Representer[Any, Any, Any, TorchLogitCategoricalDistribution]

Representer for torch-uncertainty calibration scalers that return logits.

Initialize the representer with a predictor.

Parameters:: predictor – The predictor to be used for building representations.

__call__(*args: In.args, **kwargs: In.kwargs) → R[source]¶: Alias for the represent method.

predict(*args: In.args, **kwargs: In.kwargs) → R[source]¶: Predict the representation for a given input.

predictor: Predictor[In, Out]¶

represent(*args: Any, **kwargs: Any) → TorchLogitCategoricalDistribution[source]¶: Build a representation for a given input.